Researchers who have been using Fitbit data to predict surgical outcomes have found a new way to more accurately measure patient recovery after spinal surgery.
Using machine learning techniques developed at Washington University in St. Louis' AI for Health Institute, Chengyang Lu, Fulgraph Professor in the McKelvey School of Engineering at the university, in collaboration with Jacob Greenberg, MD, assistant professor of neurosurgery in the School of Medicine, developed a way to more accurately predict recovery from lumbar spine surgery.
The results, published this month in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, found that the model outperformed previous models in predicting outcomes in spine surgery. This is important because outcomes in lumbar surgery, and many other orthopedic procedures, can vary widely depending not only on a patient's structural disease but also on differences in their physical and mental health characteristics.
Recovery after surgery is influenced by both physical and mental health before surgery. Some people worry excessively in the face of pain, which can worsen their pain and recovery. Others have physiological issues that make pain worse. If doctors can know in advance the various pitfalls patients may face, they can better plan their treatment.
“Predicting outcomes before surgery can help establish some expectations, aid in early intervention and identify high-risk factors,” said Ziqi Xu., student in the Lu lab and first author of the paper.
Previous studies predicting surgical outcomes typically used patient questionnaires administered once or twice in the clinic, capturing static slices over a period of time.
“It has not been able to capture the long-term changes in patients' physical and psychological patterns,” Xu said, adding that previous studies of training machine learning algorithms have focused on only one aspect of surgical outcomes, “ignoring the multifaceted nature inherent in post-surgical recovery.”
See the “big picture”
Researchers have used mobile health data from Fitbit devices to monitor and measure recovery and compare activity levels over time, but this study showed that combining activity data with longitudinal assessment data could more accurately predict how patients will do after surgery, Greenberg said.
“This study is a 'proof of principle' that shows multimodal machine learning can help clinicians get a more accurate 'big picture' of the interrelated factors that influence recovery. Before embarking on this study, the research team first developed statistical methods and protocols to ensure they provided the artificial intelligence system with the right balanced data.”
The team previously published a paper in Neurosurgery showing for the first time that objective patient-reported wearable measurements improve prediction of early recovery compared with traditional patient assessments. In addition to Greenberg and Xu, Madelynn Frumkin, a doctoral student studying psychology and brain sciences in the Thomas Rodebaugh lab in the School of Arts & Sciences, was co-first author of the paper. Wilson “Zack” Ray, MD, PhD, the Henry G. and Edith R. Schwartz Professor of Neurosurgery in the School of Medicine, was co-senior author with Rodebaugh and Lu. Rodebaugh is currently at the University of North Carolina at Chapel Hill.
The study showed that Fitbit data correlated with multiple surveys assessing a person's social and emotional state. The data was collected through an “ecological momentary assessment” (EMA), which uses smartphones to frequently prompt patients to assess their mood, pain levels and behavior multiple times throughout the day.
“We combine wearables, EMA and clinical records to collect a wide range of information on patients, from physical activity to subjective reports of pain and mental health, as well as clinical characteristics,” Lu said.
Greenberg added that cutting-edge statistical tools such as “dynamic structural equation modeling,” which Rodebaugh and Frumkin helped advance, were critical to analysing the complex longitudinal EMA data.
Aiming for better long-term results
In their latest study, the researchers took all these factors into account and developed a new machine learning technique called “multimodal multitask learning” to effectively combine different types of data to predict multiple recovery outcomes.
In this approach, the AI learns to assess the associations between outcomes while capturing differences in outcomes from multi-modal data, Lu added.
According to Xu, the method captures shared information about the interrelated tasks of predicting different outcomes and leverages that shared information to help the model understand how to make accurate predictions.
All of this is compiled into a final package to generate predicted changes in postoperative pain interference and physical function scores for each patient.
Greenberg said the study is ongoing so that researchers can continue to fine-tune the model, conduct more detailed evaluations, predict outcomes, and, most importantly, “understand what types of factors we might be able to modify to improve long-term outcomes.”
Xu Z, Zhang J, Greenberg JK, Frumkin M, Javeed S, Zhang JK, Benedict B, Botterbush K, Rodebaugh TL, Ray WZ, Lu C. 2024. Predicting multidimensional surgical outcomes with multimodal mobile sensing: A case study of patients undergoing lumbar spine surgery. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 8, 2, Article 81 (May 2024) https://doi.org/10.1145/3659628
Greenberg JK, Frumkin M, Xu Z, Zhang J, Javeed S, Zhang JK, Benedict B, Botterbush K, Yakdan S, Molina CA, Pennicooke BH, Hafez D, Ogunlade JI, Pallotta N, Gupta MC, Buchowski JM, Neuman B, Steinmetz M, Ghogawala Z, Kelly MP, Goodin BR, Piccirillo JF, Rodebaugh TL, Lu C, Ray WZ. Preoperative Mobile Health Data Improves Recovery Prediction from Lumbar Spine Surgery. Neurosurgery. March 29, 2024. https://doi.org/10.1227/neu.0000000000002911
The research was funded by grants from AO Spine North America, the Cervical Spine Research Association, the Scoliosis Research Association, the Barnes-Jewish Hospital Foundation, the University of Washington/BJC Healthcare Big Ideas Competition, the Fullgraf Foundation, and the National Institute of Mental Health (1F31MH124291-01A).
